- From: Peter Murray-Rust <Peter@ursus.demon.co.uk>
- Date: Sat, 08 Feb 1997 11:06:25 GMT
- To: w3c-sgml-wg@w3.org
In message <199702080118.RAA04259@boethius.eng.sun.com> bosak@atlantic-83.Eng.Sun.COM (Jon Bosak) writes: [... suggestion for XML catalogs ...] I support the _idea_ of this proposal, because it gives an impetus to creating a managed namespace. Sometimes such namespaces can be cheap to create and it simply needs some-body to make a proposal which is sufficiently convincing, and suffiently simple to implement that everyone sees the value and adopts it. (IMO MIME has some of these features though it also has design flaws.) At present the most obvious namespace management available to everyone is DNS. (For example, in 'Java in a Nutshell', p 19, O'Reilly show how it can be used to provide a unique namespace for any Java package, e.g: COM.ora.writers.david.widgets.Barchart; (The first two components are the domain name in reverse order). I am not familiar with the URN process in detail - am I right in assuming that it based on DNS? > Let's suppose that "-//FOOCORP::FOOAUTH//DOCUMENT sometitle > ver1.01//EN" fails to find resolution at the local level. Postulate > the existence of a server (probably mirrored, but let's stay above > that level of detail for the nonce) funded by every organization that > wishes to use this mechanism. Let's call this server xmlid.net. That > server does not have a listing for every FPI; all it has is one (and > only one) entry for every publisher that has joined the cooperative. ^^^^^^^^^ Jon, could you expand on this word, please? If it means 'Those large corporations whose business is publishing information and who are prepared to pay for a maintained site (xmlid.net) to be set up'. If this is the case then it will lead either to a fragmentation of the community, or to a proliferation of 'xmlid.net's (e.g. 'xmlid.net.ac.uk', 'xmlid.bio.net') > So on that server there is a single lookup table cached in RAM, one > line of which consists of FOOCORP::FOOAUTH on one side and > foodocs.foo.com on the other. This server is going to have to be funded somehow, with a registration process and removal of dead entries. Will it, in itself, be able to scale to respond to demand or will there be a DNS-like system of servers? > > Add to the abilities assumed for the XML client under the XML catalog > proposal the following: it can query xmlid.net with the string > FOOCORP::FOOAUTH and get back foodocs.foo.com; it can then query > foodocs.foo.com for its corporate catalog; and it can add that > corporate catalog to its own for long enough to resolve the query (or > fail, but that's entirely FooCorp's responsibility). > > This isn't enough to build a resolution mechanism, but it's enough to > think that building one might be possible. Is there anything > basically wrong with the idea? It seems to involve a mapping of organisations onto domain names. For example SUN:: would map onto sun.com (right?). Is SUN:: the only component of SUN or are there also SUNSOFT, SUN_MICROSYSTEMS, etc. It seems to me that it is necessary to have a feel for the stability of organisation names before you can get an idea of the work involved. If domain names are a useful approach, what about the use of something like: http://www.sun.com/xmlcatalog/ This says, in effect, if your site (or server) has implemented XML then this is reserved request for the XML catalog that relates to XML information on that site/server. It's unlikely to clash with existing namespace (apart from people who used XML for something else :-). Webmasters could easily implement it, and it should be fairly straightforward to update the local catalog. P. -- Peter Murray-Rust, (domestic net connection) Virtual School of Molecular Sciences, Nottingham University, UK http://www.ccc.nottingham.ac.uk/~pazpmr/
Received on Saturday, 8 February 1997 07:47:14 UTC